Field of the Invention
The invention generally relates to registered fingerprint datasets updating technology for fingerprint recognition, and more particularly, to a smart learning method for updating the registered fingerprint datasets.
Description of the Related Art
In recent years, biometric recognition technology has developed greatly. Since security codes and access cards may easily be stolen or lost, more attention has been paid to fingerprint-recognition technology. Fingerprints are unique and never-changing, and each person has multiple fingers for identity recognition. In addition, fingerprints can be obtained easily using fingerprint sensors. Therefore, fingerprint recognition can provide increased security and convenience, and financial security and confidential data can be better protected.
With conventional fingerprint recognition technology, when the user registers his or her finger, the registered fingerprint datasets cannot be modified. Therefore, for fingerprint recognition, the user needs to put his or her finger at the same or similar angle and location on the fingerprint sensor as or to those where the user puts the finger during fingerprint registration. If the user fails to put his or her finger at the same or similar angle and location, the fingerprint sensor will not be able to capture sufficient minutiae for the verification fingerprint data to be compared to and match any of the registered fingerprint datasets, which may cause the fingerprint recognition fail. However, the user may put the finger at different angles and/or locations when he or she puts the finger on the fingerprint sensor for different times. Therefore, during fingerprint recognition, due to the change of the angle and/or location where the user puts his or her finger on the fingerprint sensor, the fingerprint sensor may not be able to capture sufficient minutiae for comparison. As a result, the user may need to put his or her finger on the fingerprint sensor many times in order to pass the fingerprint recognition.